BiostatisticsTier 1Medical Sciences concept

Hierarchy of evidence in study design

Core concept

For a question of treatment efficacy

Level	Design
1	Systematic review / meta-analysis of RCTs
2	Individual RCT (adequately powered, low risk of bias)
3	Non-randomised controlled / cohort study
4	Case-control study
5	Case series, cross-sectional, case report
6	Expert opinion, bench/animal work, mechanistic reasoning

Randomisation is what buys causal inference - it balances unmeasured confounders, which no analytical adjustment can do
A poorly conducted meta-analysis of poor trials sits below one excellent RCT. The hierarchy ranks designs, GRADE ranks the actual evidence.

The hierarchy is question-specific

Question	Best design
Treatment efficacy	RCT / SR of RCTs
Harm, rare or delayed	Cohort, or case-control if the outcome is very rare
Prognosis	Inception cohort followed from a uniform disease point
Diagnostic accuracy	Cross-sectional study vs a reference standard, consecutive patients, blinded
Prevalence	Cross-sectional survey
Aetiology of a very rare disease	Case-control; sometimes a case series is all there is

GRADE

Starts high for RCTs, low for observational studies
Downgrade for: risk of bias, inconsistency, indirectness, imprecision, publication bias
Upgrade observational data for: large effect, dose-response, plausible confounding that would only weaken the observed effect
Parachute argument - some effects are so large that an RCT is unnecessary and unethical

Intention-to-treat preserves randomisation and is the primary analysis for superiority; per-protocol is the conservative analysis for non-inferiority
Observational data are overturned by RCTs with some regularity - HRT and coronary disease (healthy-user bias), beta-carotene, vitamin E
Registry and real-world data answer generalisability, uptake and rare harm - not efficacy
Mendelian randomisation uses genotype as a natural randomisation to strengthen causal claims from observational data (LDL, Lp(a))
N-of-1 trials sit high for an individual patient despite low external validity

3 more sections, plus exam facts

Premium unlocks every note across every specialty, and the full exam fact library behind it.